Eliminating a type instability

New to Julia
1

Hi All,

I’m getting a type instability in a generated function I’ve written to perform Clenshaw’s recursion on multivariate Chebyshev polynomials. Here is a Gist containing a minimum working example: https://gist.github.com/RJDennis/81b7d62f12878a14ef1a07a3d7b4a2e9

The type instability is in p which is an Array, but assigned type Any. Suggestions on how to eliminate this type instability would be most welcome.

Thanks,

Richard

2

Try Base.tail instead of 2:N

3

Quite why that works where 2:N doesn’t is a bit of a mystery, but it works so thank you.

4

That worked. Thanks.

5

I’m also curious why that works.

Also, Dennis, I looked at the code from your generated function (by leaving off the @generated), and you might want to look at the order of the loops. I don’t think they are the best order for Julia’s column-major ordered arrays.

6

Because the size of the tuple is unknown when you are indexing with a general unit range.